[pgpool-general: 3079] Re: pgpool -m fast problem

[Yugo Nagata]

Hi,

On Mon, 28 Jul 2014 14:01:08 +0200
Tomasz Straszewski <straszewski.tomasz at gmail.com> wrote:

> Hi,
> 
> Today I had some problem when i was testing watchdog mechanizm.
> 
> os: Centos 6.5 x64
> pgpool: pgpool-II-pg93-3.3.3-3.pgdg.x86_64.rpm
> 
> Steps:
> 
> 1. Run pgpool on server1
> 2. Run pgpool on server2
> 3. Type "pgpool -m fast" on server1 to stop pgpool
> 
> After that pgpool on server2 is escalating to master correctly and VIP 
> is bringing up. On server2 VIP is down but stop command is still working 
> and couldn't stop.
> Additionaly i can see after typing "netstat -tlnp" that pgpool process 
> is still running.
> 
> Stopping process on server1 is running until i invoke "pgpool -m fast" 
> command on server2.
> 
> Maybe some connection between server1 and server2 is not closing during 
> stop command execution?

There is a bug about VIP control in watchdog of pgpool-II 3.3.3. 
pgpool-II checks an exit code of ping when bringing up VIP, but
due to the bug, new active pgpool (on server2 in your case) goes
into an infinite loop. In addition, previous active (server1) also
hangs since this is waiting for response from the new active, which
is in hang and disable to respond.

This is fixed in this commit;
http://git.postgresql.org/gitweb?p=pgpool2.git;a=commitdiff;h=ac397a13d18d18c63cc2fbbe0e5d0e5aad845293

Could you please get patch or latest tar ball from here, and try to build
pgpool from source files? Or, please wait for next stable version 3.3.4,
which will be released at the beginning of next month.

> 
> Regards
> 
> 
> _______________________________________________
> pgpool-general mailing list
> pgpool-general at pgpool.net
> http://www.pgpool.net/mailman/listinfo/pgpool-general


-- 
Yugo Nagata <nagata at sraoss.co.jp>